This post is a part of a larger series on Google Optimize that will give you all the information you need to become a pro at testing in Optimize.
Check out other parts of the series:
Google created Optimize with the idea that anyone can use it. It’s easy to set up and creating tests in it is really simple because of their intuitive What You See Is What You Get (WYSIWYG) editor. If you never ran an A/B test before and are at the beginning of your website optimization journey, Optimize is a great tool to start with.
If you have Optimize set up on your website, you can start creating your own experiments right away. If you have yet to install it, check out our guide on how to add it to your website.
These are the steps you need to complete when creating a test and what we’ll be covering in this article:
Enter the container to which you have successfully linked a ‘’property’’, which is the word Optimize uses for website. In other words, ‘’linking a property’’ means that you successfully linked your website to Optimize and that you can now edit it in Optimize.
Your container is synonymous to your website and you will be adding experiments INTO your container. That’s way you can link ONLY ONE WEBSITE to each container, because your container is your website. But a container can contain a number of experiments- and all these experiments will be running on that one website
You can have UP TO 3 A/B TESTS running at the same time. This is the limit Optimize has in place and if you need to run more tests at the same time, you’ll have to get the paid version Optimize 360.
In our case, we successfully installed Optimize on our ‘’Testing.Agency Demo’’ website and that website has been filed into the ‘’Website 2’’ container. Notice that containers ‘’Website 1’’ and ‘’My Container’’ do not have any websites linked to them.
Because of the way the system is set up, and the fact that you can link only one website to each container, this means that the website on which you installed Optimize will be your Original (the A in A/B testing) by default. So the only thing you have to do is create your Variant B, which is the new version of the website you will be testing.
As soon as you enter the container, you’ll notice the ‘’Create Experiment’’ button to the right of the dashboard.
Once you do that, a panel will appear on the right. You will be prompted to:
- Name the experiment;
- Add the URL of the page you’d like to test. You add the URL of the website you already linked to the container, in our case this is http://testing.agency/demo
- Choose the type of test you are going to run. You’re running an A/B test, with only two variants. Multivariate testing is more complicated and requires more than two variants (A/B/n tests). You’ll usually only run multivariate tests if you have a lot of traffic.
Hit ‘’Create’’ and your experiment will appear on your dashboard. We named ours ‘’Button Color’’ because we plan on changing the color of our Call-To- Action button on our landing page. Since it’s still not set up fully, it will appear under the ‘’Draft’’ section of your container dashboard.
Now you can click into your experiment and get to the most fun part- setting up the new variant and making changes to your website. Once you click into it, this what your Experiment dashboard is going to show you:
- how many variants there are in your experiment,
- the percentage of traffic each variant receives,
- options that allow you to preview the way the experiment will look on mobile and desktop platforms,
- the number of edits for each variation,
- extra options such as Delete and Edit variant.
For now, you only have your Original and you can see that it’s receiving a 100% of your traffic. The goal is now to set up your first Variant (the B in A/B testing).
If you look to the right of your Dashboard, you’ll see this panel:
In our previous article, we linked our website to our container through Google Analytics. If you did this correctly, you don’t have to bother with connecting your variants to your website because you already did it. The Google Analytics code is already there and you are ready to go.
You already have the Original version of your website on the dashboard. And now you need to create the alternative version of your original, which you are testing. To do that, click on ‘’New Variant’’.
Optimize will prompt you to name your variant. We’re naming it ‘’Variant 1’’.
Click ‘’Add’’ and your variant will be on your dashboard. Now you can see that Optimize automatically splits your traffic. 50% is going to your Original (Variant A) and 50% is going to your Variant 1 (Variant B).
In addition to that, you can see how many changes have been made to Variant 1. Right now that number is 0. But that’s about to change!
Now for the fun part! Click on ‘’Variant 1’’. If you’re working in Chrome (which we recommend because Optimize is Google’s product) and you never used Optimize’s Visual editor before, you’ll be prompted to install a plug-in. It shouldn’t take you more than a few minutes to finish this step and it’s fairly straight-forward.
You install it like any other plug-in, no need to restart your browser.
And then you’ll be seeing your website in the WYSIWYG editor! This is what our website looked like.
This is still what your website looks like- you’re seeing your Original version here. What we didn’t like about it is the Call-To-Action button that was green and wasn’t that visible. So we wanted to change the button color.
Whatever element you want to change on your website, you just have to click on it and the panel to the right, ‘’Edit Element’’, will tell you which changes are possible. We selected the button and the panel showed us what we can change.
This is the part where you should play around with your website. Don’t be afraid to change things dramatically (besides, big changes are the ones that will make the biggest difference). You can easily reverse any changes you make and your test won’t go live until you give the final approval.
For the sake of this guide, we just changed the color of the button from green to orange and you can see that change here:
The editor offers you a chance to see what your website will look like on different devices. Just click on the device type at the top of the screen and choose your device.
Once you’re done making all the changes you want to test with your variant, click ‘’Save’’ and ‘’Done’’ and you’ll be taken back to your experiment dashboard.
With targeting you can pick a very specific portion of your audience and expose only that portion to your experiment. You can choose only to target people from a specific city or those using a certain browser. Or you can have very general settings and just choose to have all your visitors participate in the A/B test.
There on your Dashboard, you’ll see that your experiment is still a draft and that there are some steps you need to complete before your experiment can go live.
If you click on ‘’Show Steps’’, it will tell you what other steps you need to complete before the ‘’Start Experiment’’ button becomes available.
‘’Targeting’’ is marked as completed because it’s done automatically. If you click on the ‘’Targeting’’ tab, right next to ‘’Objectives’’, you’ll see the default settings. For your first test, you can leave the targeting settings as they are, but here’s a quick overview of what each setting enables you to do.
‘’Percentage of visitors to target’’ defines the percentage of your overall traffic that will be affected by the test. Here, all of you visitors will be participating in the test because it’s set to 100%. What this setting is telling you is that, everyone who visits the website is either going to see Variant A or Variant B. If you set it to 65%, then this means that only 65% of your overall traffic is going to be taken into account and only 65% might have the chance of being directed to Variant B. The other 35% of your traffic will ONLY be exposed to Variant A, at all times.
This is what Google recommends:
‘’This controls how many people who visit your site are included in the experiment. Everyone else will see your original page. If you want quicker results, you might want to include a higher proportion of your visitors in the experiment. However, if your experiment is drastic and risky, you might want to include only a small proportion of your site’s visitors.’’
The ‘’Weighting of visitors to target’’ shows you that 50% of your traffic will be going to Variant A and 50% will be going to Variant B. If you want to change this, just click ‘’Edit’’ and adjust the traffic. Note: For statistical significance, we highly recommend you always direct an equal amount of your traffic to each variant.
The next section in Targeting will define when you want the test to start. By default, the test will start as soon as Variant 1 loads. You can check and see that the page which you changed in the Editor is actually the one defined in the URL field.
Otherwise, if you’d like to specify your targeting even more, you can click on ‘’AND’’ which will give you even more targeting filters.
With the help of these rules, you can set up your test so that, for example, only people using a certain technology or coming from a certain city or country are exposed to the test. Depending on your needs, your test can target a very specific, narrow audience. But, you can also leave the targeting settings Optimize chooses by default and you’ll still be able to run a valid test.
Objectives help you define what you want to test with these changes. The free version of Optimize lets you choose up to 3 different objectives (Primary objective and two Secondary objectives). The objectives at your disposal are: Pageviews, Session Duration, Bounces. If you need more than 3 objectives, you’ll need to upgrade to the paid version, Optimize 360.
With this experiment our primary objective is to track bounces or the percentage of visitors who don’t browse any further than the landing page. The secondary objective is to track session duration or amount of time a visitor spends actively using our website.
In the ‘’Description and hypothesis’’ field you can write out the idea behind the test and what results you want to achieve. For a simple test like this, we can write something basic like:
‘’Because of a high bounce rate and short average session duration, we changed the color of our CTA button on our landing page. This should lower our bounce rate and visitors will browse our website further, rather than abandoning the website directly from the landing page.’’
Notice how your experiment is still a draft and Optimize is telling you that there are steps you need to complete before you can run your test. Once you’re done setting up your objectives, click the ‘’Save’’ button in the upper right corner.
As soon as you click ‘’Save’’, the message will change and you can run your first experiment!
Optimize will advise you to check the pages where you want to run your test when you click on ‘’Start Experiment’’. It’s essentially telling you to go and check that your experiment is actually running properly and being displayed accurately on your website AFTER you click ‘’Start’’.
Hit ‘’Start’’ and YOUR FIRST TEST IS LIVE!
Remember to access your website from different devices and browsers to make sure your Variant 1 is being displayed properly.
Since Optimize is so well integrated with Google Analytics, you don’t have to worry about how you’re going to interpret the data. Sit back and once your test is done, just go to the ‘’View report in Google Analytics’’ link.